I recently built a drone from the ground up - learned how to build PCB's with the ESP32, wrote all of the flight firmware, etc. and built a controller iOS app.
Extended Kalman Filters are even more interesting because they let you do sensor fusion and such
Why is it exactly, that mobile providers can’t catch this…?
Just like with SMS fraud. It might cost them a few cents per subscriber to do effective anti-spam measures, but now society has to pay the cost
The MVNOs such as MobileX mentioned above do not have their own towers or the cellular backbone i.e. the core network. They'll merely use the MNO such as Verizon or T-Mobile and have a commercial arrangement in which MVNO just handles the marketing and the customer support. So the MVNOs may not have the right tools or data or incentives to catch these sim farms.
It depends on how closely they monitor the network and how much these people abused it. A steady, slow increase in calls/messages on a site wouldn’t show up in stats unless there was a lot of constant congestion, and even then most telcos these days outsource a lot of their network monitoring and capacity management to contractors that just don’t care.
Plus if they’re using legacy 2G/3G, it’s not the shiny thing that most telco network quality crews care about for customer experience…
I am not an expert in AI by any means but I think I know enough about it to comment on one thing: there was an interesting paper not too long ago that showed if you train a randomly-initialized model from scratch on questions, like a bank of physics questions & answers, models will end up with much higher quality if you teach it the simple physics questions first, and then move up to more complex physics questions. This shows that in some ways, these large language models really do learn like we do.
I think the next steps will be more along this vain of thinking. Treating all training data the same is a mistake. Some data is significantly more valuable to developing an intelligent model than most other training data, even when you pass quality filters. I think we need to revisit how we 'train' these models in the first place, and come up with a more intelligent/interactive system of doing so
From my personal experience training models this is only true when the parameter count is a limiting factor. When the model is past a certain size, it doesn't really lead to much improvement to use curriculum learning. I believe most research also applies it only to small models (e.g. Phi)
Wow. I really like this take. I've seen how time and time again nature follows the Pareto principle. It makes sense that training data would follow this principle as well.
Further that the order of training matters is novel to me and seems so obvious in hindsight.
Maybe both of these points are common knowledge/practice among current leading LLM builders. I don't build LLMs, I build on and with them, so I don't know.
This is precisely why chain of thought worked. Written thoughts in plain English is a much higher SNR encoding of the human brain's inner workings than random pages scraped from Amazon. We just want the model to recover the brain, not Amazon's frontend web framework.
Apple’s biggest problem is their commitment to privacy. Delivering effective AI requires a substantial amount of user data that Apple doesn’t collect.
Their other problem is they value designers and product managers more than engineers (especially top tier AI engineers).
Both problems are basically the death knell of any hope for Apple to have good AI, but combined? It’s never gonna happen. Which is sad because Apple’s on-device hardware is quite good.
Thanks to the EU for ruining the web by forcing everyone to show the ridiculous "Accept Cookies!" agreement. No wonder people prefer native apps. They’re better - for a lot of reasons, both because they can interface more cleanly with OS specific features and also for performance.
And 'privacy' is a horrible argument to prefer websites over apps. For the average person (not a privacy obsessed techie) - the web is just as bad if not worse from a privacy perspective than native apps.
I do agree that not everything needs an app - websites have their place. But when I go to browse HN on my phone, I don't do it through the web, I do it through Octal (which is open source).
Frankly I am tired of privacy-obsessed techies ruining tech for everyone else. Let's face it - 99% of the things you're worried about are simply going to let companies....show you ads that are more relevant to your life. The horror!
I think you are right here - the ability to test theory of mind in an LLM would be more like testing how well it can distinguish its own motivations/ideas from that of a separate entity.
I would agree that this question is more of a logic puzzle and less of a real test of 'theory of mind'
In fact, just to have a theory of mind, it kind of assumes you have a mind, with your own ideas/motivations/etc
Yep, there is a big reason why Europe has so few successful big tech companies, it is a regulatory hellscape. They have so many pointless privacy regulations that only the “big” companies can even hope to compete in many markets like ad tech.
There is a big reason why the USA outside of Silicon Valley and Seattle has so few successful big tech companies: because success begets success and capital breeds more capital. If it was just European regulation you'd expect SV equivalents everywhere except for Europe. That didn't happen.
And the last thing we need is more competition in ad tech.
Given the size and abuses by the existing ad tech giants, why do you say they last thing we need is more competition. wouldn't more competition mean they have less money and get away with less and have to behave better?
No, we need them to go away. Like the dinosaurs or the Dodo. Because competition between ad tech giants means the public is collateral damage in the ensuing war. That's because the advertiser money flows to the ad tech company that is most efficient at extracting dollars from their audience. Even a .1% increase is enough to swing the battle, and that arms race has been running since 1994 or so, the results are there for all to see.
As much as we would like them to magically disappear and get uninvented, like NFTs, there doesn't seem to be any mechanism by which that realistically happens. So then, like harm reduction, isn't more competition better than less? It means less money goes to the existing giants, which may not totally starve them, but will put them on a diet.
Silicon Valley equivalents are brimming up in other parts of the world, Taiwan is very much known for its hardware technology. There are documentaries about Shenzhen becoming a tech hub too. Even here in Bangalore (India), there are many tech companies doing massive amount of good work.
But they're also right in the sense that regulation acts like a barrier in many parts of the world. I had often wondered why did Linus Torvalds and other Engineers travelled to Silicon Valley to found Linux, etc? Did they not find opportunity in Finland or any other nearby European countries?
That's because once you have a runaway success the US will tax you lower and your quality of life will be higher than what you can achieve in Europe. The USA is a great country if you're on to a winner. So the vacuum cleaner in SV tends to suck the air out of a lot of EU successful start-ups and engineering efforts simply because that's where the money is. Typical start-up valuations in the USA dwarf those in Europe, access to a single unified and mostly mono-lingual market are far more of a factor than any EU regulations, that's just a dumb meme that gets tossed around by the clueless. Yes, taxes are higher. But so is average quality of life, as opposed to average GDP, which relies on outliers.
Despite being less wealthy and diverse in both language and culture, China and India produce more tech unicorns than the EU. Several smaller countries, like Singapore and Israel also do far better than the EU on a per-capita basis.
I'd attribute most of the gap to regulatory and cultural difference.
EU is also not uniform in terms of language - so most EU companies need to decide whether they go global and start in a foreign (English) language or begin in their native language and risk getting locked in there.
I’ve been mentoring startups in EU for over 10 years and there were only a handful that had issues with regulation, but 95% had issues with a language/country lock in.
You missed the per-capita examples of Israel and Singapore. If either had the population of Germany or even that of Spain, they'd have more unicorns than the entire EU.
It's not striking that the EU is as wealthy per-person as it is and has so few tech unicorns. It's also not striking for a region with hundreds of millions of people. What's striking is that despite being wealthy and populous, the region hasn't done too well with tech.
I'd say that it's even changed during my own lifetime. There was a time when German cars had a much larger market share and Nokia was a dominant phone company. Nokia failed the transition into the smartphone era and while German cars are still great, their market share in EVs is much, much smaller. And it's not like there's a lack of talent. Plenty of Europeans are building huge tech companies, but a large fraction are choosing to do so in the US or other similar markets, like Canada (e.g., Shopify).
GDP always gets trotted out as if it is a holy grail and a benchmark for social welfare, which it isn't so think of it as a (failed) preemptive strike against getting a longer comment thread.
And so have Eindhoven (ASML), London (Revolut, Monzo, Wise and Deliveroo), Paris (DailyMotion, AppGratis), Berlin (Soundcloud, Mister Spex, Zalando, Helpling, Delivery Hero, Home24 and HelloFresh) and Amsterdam (Sonos (ok, technically Hilversum), Booking.com, TomTom) etc, etc. So what?
Tech companies exist the world over. The specific kind of tech company that requires a mountain of free cash and that can monopolize a whole segment is a SV anomaly and Microsoft is the exception simply because of when it started.
Those are actually tiny compared to most US tech companies with a global reach. The issue is that here in Europe everyone speaks their own language, and it's not feasible to advertise your tech stuff to the entire continent. There's no TV channels here where you can reach 300M people with a single commercial in a single language.
It's also an issue with capital. Everyone was shocked when Mistral raised what, 300M dollars? Ask on the street if anyone's heard of Mistral, and then ask about ChatGPT.
Meanwhile effing xAI from Elon, that no one really cares about is looking to raise $1B.
Here in Europe we're sadly not on the same level. Available capital is smaller. Reach is smaller (in practice but not in theory). Profit margins are smaller. Regulation is higher.
In 2023 you need extreme luck to create something in Europe that reaches a global audience to the point it's not worth trying. Just go for your local domestic market instead.
> Those are actually tiny compared to most US tech companies with a global reach
What are you talking about? Booking is 5.3B revenue, 112B market cap. Adyen is 37B market cap. These are not "tiny" companies compared to public tech companies in the US, and there are more than just these two.
Sure, Europe doesn't have as many frothy VC's and associated tech companies with insane valuations as the US. But it's not trailing out in last place like some of these comments make it out to be.
People need to look at actual facts and numbers before regurgitating the same old memes about how terrible Europe and the EU are.
reply